RC_N

We are releasing RC_N to FireCloud production on 2019-01-16.

This release includes the following Leo issues:
RC_N_2018-01-16

• Retry 409s from GoogleIamDAO (#738)
  • Fixes a reliability problem leading to cluster errors when many clusters are concurrently created in the same project.
• Install the table of contents notebook extension by default (#404)
  • User facing: enables Jupyter table of contents by default. This is an unobtrusive and commonly used extension.
• Upgrade FireCloud UI to the latest Leo Jupyter image (#699)
  • User facing: Updates FC to use the latest image including python 3.6, R 3.5, and Hail 0.2
• Separate front- and back- Leo functionalities (#564)
  • Not user facing: prepares for expanding Leo to run on multiple instances.

RC_K

We are releasing RC_K to FireCloud production on 2019-01-03.

This release includes the following Leo issues:

• https://github.com/DataBiosphere/leonardo/labels/RC_K_2018-01-03

Tickets in this release

Bug Fixes

• #710: Some R packages are failing to install due to a system dependency
  • User facing: Some R packages were failing to install due to a system dependency. This has now been fixed.

Enhancements

• #542: Make cluster scopes definable by app/user.

  • User facing: Apps/Users can specify there own scopes for the cluster.

• #713: Add new version end point for Leonardo. Userfacing.

Infrastructure

• #562: Refactor cluster monitoring.

  • Part of horizontal scaling. Not user facing.

• #687: Make cluster patching horizontally scalable.

  • Part of horizontal scaling. Not user facing.

• #695: Add rstudio base image to Leonardo code. Update the CI process to accommodate the change.

  • Not user facing.

• #707: Add jitter to prevent timeout when claiming project from gpalloc.

  • Not user facing.

RC_K

We are releasing RC_K to FireCloud production on 2018-12-04.

This release includes the following Leo issues:

• RC_J_2018-11-27
  and
• RC_K_2018-12-04

Tickets in this release:

Bug fixes

• #670: Prevent kernel disconnections upon idling for an hour or more
  • User facing: fixes a bug which caused a kernel disconnect after 1 hour.
• #486: 500 Error when downloading notebook as ipynb
  • User facing: Jupyter UI no longer throws a 500 error when downloading .ipynb files
• #474: Direct JupyterLab notebook URLs don't work with Leo
  • User facing: Direct links to notebooks now work in JupyterLab.
  • Note: JupyterLab will now be installed by default on Leo clusters as of this release. It can be accessed by appending /lab to the proxy URL. For example: https://notebooks.firecloud.org/notebooks/my-project/my-cluster/lab.

Enhancements

• #692: Allow users to pick their Jupyter image at cluster creation time
  • User facing: adds a new optional field jupyterDockerImage to the createCluster API to control the Jupyter docker image which gets installed on the clusters. Defaults to the latest prod image.
• #580: Docker image updates. All user facing!
  • #663: Upgrades python from 3.4 -> 3.6
  • #639: Upgrades R from 3.4 -> 3.5
  • #488 Upgrades panda from 0.22.0 -> 0.23.2
  • #407: Upgrades the Jupyter image to be based on debian stretch instead of jessie
  • #193: Upgrades Hail from 0.1 -> 0.2
  • #144: Upgrades Dataproc from 1.1 to 1.2
• #496: Allow changing auto-pause threshold for Running clusters
  • User facing: A cluster's auto-pause settings can now be updated dynamically for running clusters via the Leo updateCluster endpoint.
• #288: Add ability to add/remove worker nodes to a running cluster
  • User facing: A cluster can now be resized using the Leo updateCluster endpoint. It currently supports add/removing worker and preemptible nodes.

Infrastructure

• #677 - Monitoring Cluster status: ClusterName(xyz) Error was not equal to Running - Add retries to the docker commands in init-actions
  • User facing: Makes cluster creation more resilient to transient errors
• #657: Cluster cleanup script erroring due to Cloud Dataproc API not being enabled
  • Not user facing: Fixes an issue with a script to cleanup dev clusters
• #563: Refactor DNS caching
  • Not user facing: Changes to the Leo proxy to allow for better horizontal scaling of instances.
• #148: Leo: pin all Jupyter docker dependencies
  • Not user facing: Add more control of Jupyter image dependencies to prevent unexpected breakages.

RC_J

We are releasing RC_J to FireCloud production on 2018-11-27.

This release includes the following Leo issues:
RC_J_2018-11-27

• #677: Monitoring Cluster status: ClusterName(xyz) Error was not equal to Running - Add retries to the docker commands in init-actions
  • Testing only, not user facing
• #657: Cluster cleanup script erroring due to Cloud Dataproc API not being enabled
  • Not user facing
• #563: Refactor DNS caching
  • Preparation for horizontal scaling, not user facing

RC_I

We are releasing RC_I to FireCloud production on 2018-11-14.

This release includes the following Leo issues:
RC_I_2018-11-14

• #661: Add Terra to CSP
  • This allows Leo notebooks to be used from the https://app.terra.bio domain.

RC_H

We are releasing RC_H to FireCloud production on 2018-11-01.

This release includes the following Leo issues:
RC_H_2018-11-01

• #596: Return 403 instead of 401 for authentication errors
  • Fixes a Leo response code, not really user facing
• #589: Don't allow deleting a cluster in Creating status
  • User facing: Leo now does not allow deleting a cluster in Creating status
• #650: Add AoU domains to CSP
  • AoU facing: allows AoU app to render Leo notebooks in an iframe

RC_G

We are releasing RC_G to FireCloud production on 2018-10-23.

This release includes the following Leo issues:
RC_G_2018-10-23

• #625: Increase dataproc init script timeout
  • (User facing) Increases the dataproc init script timeout from 10 to 30 minutes to support long-running user scripts. The value is now configurable in Leonardo.
• #633: Return 204 on success for setCookie endpoint
  • (User facing) The setCookie endpoint now returns 204 on success instead of 200.

RC_F

We are releasing RC_F to FireCloud production on 2018-10-11.

This release includes the following Leo issues:
RC_F_2018-10-11

The theme of this release is performance/reliability improvements based on load testing.

• #608: Repeated frequent Ajax calls by the UI on an idle page
  • User facing: Fixes a FireCloud UI bug where repeated /localize AJAX calls were incorrectly being made to a Stopped cluster
• #624: Added ability to list clusters within a project
  • User facing: Adds a new Leo endpoint to list clusters belonging to a given Google project. This performs significantly better than the existing listClusters endpoint which lists clusters across all projects the user has access to. FireCloud UI has been updated to call this endpoint from the workspace Notebooks page.
• #622: Increase retries for init bucket creation
  • Not user facing: Addresses a failure mode when many clusters are created concurrently.

RC_C

We are releasing RC_C to FireCloud production on 2018-09-25.

This release includes the following Leo issues:
RC_C_2018-09-25 :

User facing FireCloud UI bug fixes:

• Cluster creation errors no longer cause clusters to appear to get stuck in Creating status in the UI (#590, #582)
• The cluster creation dialog now includes static text informing users of the expected time to create a cluster (#583)
• Users should no longer see 404 errors when opening a notebook from FireCloud (#577)
• FireCloud now displays appropriate error messages when a user does not have permission to modify notebooks (#572)

User facing Leonardo changes:

• Enable auto pause of clusters after 30 minutes of idle time (#585)
• Leonardo now better handles situations when clusters or projects are deleted in Google (#494)
• Leonardo is now more resilient to timeouts from GCR when pulling the notebook image (#570)

RC_B

We are releasing RC_B to FireCloud production on 2018-09-18.

This release includes the following Leo issues:
RC_B_2018-09-18 :

• #434: Add R function to Leonardo to configure bigrquery credential.
  • User facing: we now provide a library encapsulating a number of functions researchers might user in R from a notebook. The repo & documentation can be found here: https://github.com/DataBiosphere/ronaldo
• #559: Update gpalloc to scrub clusters from billing projects
  • Not user facing: updates test infrastructure (gpalloc) to remove dataproc clusters from test projects.